Improve training code 2

tests/test_onboarding_flow.sh

@@ -622,7 +622,7 @@ fi
 echo ""
 
 # =================================================================
-# STEP 4: MODEL TRAINING (ONBOARDING PAGE STEP 4)
+# STEP 4: MODEL TRAINING (ONBOARDING PAGE STEP 4) - FIXED
 # =================================================================
 
 echo -e "${STEP_ICONS[3]} ${PURPLE}STEP 4: AI MODEL TRAINING${NC}"
@@ -633,22 +633,28 @@ log_step "4.1. Starting model training process with real data products"
 
 # Get unique products from the imported data for training
 # Extract some real product names from the CSV for training
-REAL_PRODUCTS=$(tail -n +2 "$PREPARED_CSV" | cut -d',' -f2 | sort | uniq | head -3 | tr '\n' ',' | sed 's/,$//')
+REAL_PRODUCTS_RAW=$(tail -n +2 "$PREPARED_CSV" | cut -d',' -f2 | sort | uniq | head -3 | tr '\n' ',' | sed 's/,$//')
 
-if [ -z "$REAL_PRODUCTS" ]; then
+if [ -z "$REAL_PRODUCTS_RAW" ]; then
     # Fallback to default products if extraction fails
-    REAL_PRODUCTS='"Pan de molde","Croissants","Magdalenas"'
+    REAL_PRODUCTS_ARRAY='["Pan de molde","Croissants","Magdalenas"]'
     log_warning "Could not extract real product names, using defaults"
 else
-    # Format for JSON array
-    REAL_PRODUCTS=$(echo "$REAL_PRODUCTS" | sed 's/,/","/g' | sed 's/^/"/' | sed 's/$/"/')
-    log_success "Extracted real products for training: $REAL_PRODUCTS"
+    # Format for JSON array properly
+    REAL_PRODUCTS_ARRAY='['$(echo "$REAL_PRODUCTS_RAW" | sed 's/,/","/g' | sed 's/^/"/' | sed 's/$/"/')']'
+    log_success "Extracted real products for training: $REAL_PRODUCTS_ARRAY"
 fi
 
-# Training request with real products
+# ✅ FIXED: Training request with correct data types matching TrainingJobRequest schema
 TRAINING_DATA="{
-    \"tenant_id\": \"$TENANT_ID\"
-    }
+    \"products\": $REAL_PRODUCTS_ARRAY,
+    \"max_workers\": 4,
+    \"seasonality_mode\": \"additive\",
+    \"daily_seasonality\": true,
+    \"weekly_seasonality\": true,
+    \"yearly_seasonality\": true,
+    \"force_retrain\": false,
+    \"parallel_training\": true
 }"
 
 echo "Training Request:"
@@ -668,15 +674,54 @@ echo "Training HTTP Status Code: $HTTP_CODE"
 echo "Training Response:"
 echo "$TRAINING_RESPONSE" | python3 -m json.tool 2>/dev/null || echo "$TRAINING_RESPONSE"
 
-TRAINING_TASK_ID=$(extract_json_field "$TRAINING_RESPONSE" "job_id")
-if [ -z "$TRAINING_TASK_ID" ]; then
-    TRAINING_TASK_ID=$(extract_json_field "$TRAINING_RESPONSE" "id")
-fi
-
-if [ -n "$TRAINING_TASK_ID" ]; then
-    log_success "Training started successfully - Task ID: $TRAINING_TASK_ID"
+# ✅ FIXED: Better error handling for 422 responses
+if [ "$HTTP_CODE" = "422" ]; then
+    log_error "Training request failed with validation error (HTTP 422)"
+    echo "This usually means the request doesn't match the expected schema."
+    echo "Common causes:"
+    echo "  - Wrong data types (string instead of integer)"
+    echo "  - Invalid field values (seasonality_mode must be 'additive' or 'multiplicative')"
+    echo "  - Missing required headers"
+    echo ""
+    echo "Response details:"
+    echo "$TRAINING_RESPONSE"
+    
+    # Try a minimal request that should work
+    log_step "4.2. Attempting minimal training request as fallback"
+    
+    MINIMAL_TRAINING_DATA='{"seasonality_mode": "additive"}'
+    
+    FALLBACK_RESPONSE=$(curl -s -w "\nHTTP_CODE:%{http_code}" -X POST "$API_BASE/api/v1/tenants/$TENANT_ID/training/jobs" \
+        -H "Content-Type: application/json" \
+        -H "Authorization: Bearer $ACCESS_TOKEN" \
+        -H "X-Tenant-ID: $TENANT_ID" \
+        -d "$MINIMAL_TRAINING_DATA")
+    
+    FALLBACK_HTTP_CODE=$(echo "$FALLBACK_RESPONSE" | grep "HTTP_CODE:" | cut -d: -f2)
+    FALLBACK_RESPONSE=$(echo "$FALLBACK_RESPONSE" | sed '/HTTP_CODE:/d')
+    
+    echo "Fallback HTTP Status Code: $FALLBACK_HTTP_CODE"
+    echo "Fallback Response:"
+    echo "$FALLBACK_RESPONSE" | python3 -m json.tool 2>/dev/null || echo "$FALLBACK_RESPONSE"
+    
+    if [ "$FALLBACK_HTTP_CODE" = "200" ] || [ "$FALLBACK_HTTP_CODE" = "201" ]; then
+        log_success "Minimal training request succeeded"
+        TRAINING_TASK_ID=$(extract_json_field "$FALLBACK_RESPONSE" "job_id")
+    else
+        log_error "Both training requests failed"
+    fi
 else
-    log_warning "Could not start training - task ID not found"
+    # Original success handling
+    TRAINING_TASK_ID=$(extract_json_field "$TRAINING_RESPONSE" "job_id")
+    if [ -z "$TRAINING_TASK_ID" ]; then
+        TRAINING_TASK_ID=$(extract_json_field "$TRAINING_RESPONSE" "id")
+    fi
+
+    if [ -n "$TRAINING_TASK_ID" ]; then
+        log_success "Training started successfully - Task ID: $TRAINING_TASK_ID"
+    else
+        log_warning "Could not start training - task ID not found"
+    fi
 fi
 
 echo ""